What the work actually involves

You receive batches of short Russian audio clips. Some are human speech, some are model-generated, and you usually will not be told which. Your job is to score nativeness and fluency against a rubric the client supplies, then explain the score in English: which vowel was reduced wrongly, where the stress landed on the wrong syllable, whether the intonation contour of a question sounded like a statement, whether the prosody was flat in a way that reads as synthetic rather than merely as a monotone speaker. Expect to write a few sentences of specific, citable justification per clip — not "sounds unnatural," but the phonetic or prosodic feature that gives it away.

Because Russian is spoken natively across a wide dialectal and regional range, a recurring judgment call is separating regional accent, non-standard-but-native speech, and a second-language accent from artifacts of speech synthesis. Reviewers who collapse all of these into "not native" produce noisy data. The other recurring call is consistency: your rating on clip 200 should mean the same thing as your rating on clip 3.

What micro1's screen looks for

micro1 runs an AI-led interview. It will probe your Russian at native depth — reduction, stress, palatalization, intonation patterns — and check that you can describe those phenomena in intelligible English, since all written output is in English. Expect follow-ups that push past a first answer: if you say a clip sounded off, it will ask which segment and why. Stated preferences for linguistics, phonetics, voice work, or audio engineering are preferences, not gates; demonstrated ability to articulate fine-grained differences carries more weight than the credential itself.

Logistics

  • Fully remote, contractor engagement, no fixed schedule; work is claimed from available task batches
  • Asynchronous, with project coordinators reachable for rubric clarification — raising ambiguity is expected rather than penalized
  • Volume fluctuates with client demand; treat it as supplementary rather than guaranteed weekly hours
  • Requires quiet listening conditions and decent headphones — laptop speakers are not sufficient for judging fine prosodic detail
  • Pay band of $30–65/hr reflects observed rates for this role family and varies with project and assessed depth